You are on page 1of 15

REVIEWS

M O D E S O F T R A N S C R I P T I O N A L R E G U L AT I O N

RNA-mediated epigenetic regulation


of gene expression
Daniel Holoch and Danesh Moazed
Abstract | Diverse classes of RNA, ranging from small to long non-coding RNAs, have
emerged as key regulators of gene expression, genome stability and defence against
foreign genetic elements. Small RNAs modify chromatin structure and silence
transcription by guiding Argonaute-containing complexes to complementary
nascent RNA scaffolds and then mediating the recruitment of histone and DNA
methyltransferases. In addition, recent advances suggest that chromatin-associated long
non-coding RNA scaffolds also recruit chromatin-modifying complexes independently
of small RNAs. These co‑transcriptional silencing mechanisms form powerful RNA
surveillance systems that detect and silence inappropriate transcription events, and
provide a memory of these events via self-reinforcing epigenetic loops.

RNA interference
In many organisms, intergenic or antisense transcrip- X chromosome in male flies, ultimately leading to
(RNAi). Broadly refers to RNA tion gives rise to different classes of small RNAs and increased transcription 19–21. More recently, a large
silencing pathways that use long non-coding RNAs (lncRNAs) that have emerged number of other lncRNAs have been shown to act at
Argonaute and PIWI proteins as key regulators of chromatin structure in eukaryotic a gene-specific level, rather than at a chromosomal
and small RNAs to silence gene
cells1,2. In addition to their roles in RNA degradation level, to either activate or silence transcription18,22 (see
expression.
and translational repression, small RNAs modify chro- Supplementary information S1 (figure)).
Small interfering RNAs matin and target gene expression via RNA interference A unifying mechanism by which small RNAs and
(siRNAs). 22–24‑nucleotide (RNAi) pathways3–11 (BOX 1). In many instances, nuclear lncRNAs modify chromatin structure and silence tran-
small RNAs that are generated RNAi pathways mediate histone or DNA methylation scription is the formation of RNA scaffolds. Although
from longer double-stranded
RNA precursors by the
events that repress transcription. Studies of the mustard the machineries that use RNA scaffolds have greatly
ribonuclease Dicer. plant Arabidopsis thaliana6–9,12 first demonstrated that diverged throughout evolution, a number of key simi-
post-transcriptional gene silencing, and the accompanying larities enable us to define the common themes and
DNA methylation of target loci, correlated with the pro- principles that are conserved in eukaryotes from fission
duction of small interfering RNAs (siRNAs) and thus linked yeast to mammals.
RNA-directed DNA methylation to the RNAi pathway, In this Review, we discuss recent progress in our
which had previously been described in Caenorhabditis understanding of the role of RNA in genome regulation,
elegans13. Studies in the fission yeast Schizosaccharomyces focusing on the roles of different classes of chromatin-
pombe, the ciliate protozoa Tetrahymena thermophila, bound RNAs as scaffolds for chromatin-modifying
as well as in animal germline and somatic cells, then complexes. Moreover, we review recent mechanistic
revealed a general role for RNAi and related mechanisms insights into how small-RNA amplification loops are
Howard Hughes Medical in heterochromatin formation or DNA methylation10,14–17 coupled to histone or DNA methylation to form self-
Institute, Department of Cell (see Supplementary information S1 (figure)). reinforcing positive feedback systems that maintain
Biology, Harvard Medical
RNA also regulates chromatin modifications and epigenetic states. First, we discuss how small RNAs and
School, 240 Longwood
Avenue, Boston, structure through pathways that do not involve RNAi; Argonaute (AGO) complexes are assembled and how they
Massachusetts 02115, USA. some lncRNAs, and even some mRNAs, seem to contain target specific chromatin regions for silencing, focusing
Correspondence to D.M.  signals that recruit chromatin-modifying complexes on the better understood S. pombe and A. thaliana sys-
e-mail: danesh@hms.harvard. independently of small RNAs18. Early examples include tems where distinct mechanisms have been elucidated
edu
doi:10.1038/nrg3863
X inactive specific transcript (XIST), which coats the by which siRNAs and histone or DNA methylation
Published online entire inactive X chromosome in female mammals, events form self-reinforcing epigenetic loops. Second,
2 January 2015 and RNA on the X 1 (roX1) and roX2, which coat the we review nuclear small-RNA silencing pathways in

NATURE REVIEWS | GENETICS VOLUME 16 | FEBRUARY 2015 | 71

© 2015 Macmillan Publishers Limited. All rights reserved


REVIEWS

Heterochromatin other model systems, including C. elegans, Drosophila  results, it was suggested that RNAi has an important
Regions of chromatin that melanogaster and mammals, highlighting conservation role in the initiation of heterochromatin formation
retain the condensed and divergence in the roles of these pathways in gene and and its subsequent maintenance at pericentromeric
appearance of mitotic genome regulation. Finally, we discuss the mechanisms DNA repeats via the Rdp1‑mediated recruitment of
chromosomes throughout the
cell cycle. Heterochromatic
by which lncRNAs and mRNAs interact with RNA H3K9 methylation. However, later studies indicated
regions are associated processing and chromatin-modifying machineries that a small-RNA-programmed Ago1 complex medi-
with repressive histone independently of RNAi pathways. ates the targeting of specific chromosome regions and
modifications and recruitment of H3K9 methylation23 (see below).
structural proteins, and are
RNAi-mediated heterochromatin assembly
transcriptionally silent.
RNAi-mediated transcriptional gene silencing is best The RITS complex and the emergence of the ‘nascent
Argonaute understood in S. pombe, in which many basic principles transcript’ model. A physical connection between the
(AGO). A family of proteins of the pathway were first deciphered. S. pombe contains RNAi pathway and heterochromatin was established by
that bind to small RNAs a single gene each for Ago (ago1+), Dicer (dcr1+) and the purification of Chp1, a chromodomain protein that is
and that are conserved in all
domains of life. They mediate
RNA-dependent RNA polymerase (RdRP; rdp1+) (BOX 2). required for silencing the same heterochromatic regions
target recognition via Deletion of any of these genes was shown to result in loss targeted by RNAi23,24. Chp1 was found to be a component
base-pairing interactions of heterochromatic gene silencing at pericentromeric DNA of an RNA-induced transcriptional silencing (RITS) complex
between their bound small repeat regions and a reduction in the levels of histone that also contains Ago1 and heterochromatic small RNAs
RNA and complementary
H3 lysine 9 (H3K9) methylation, a conserved marker (BOX 2). The third RITS complex subunit, Tas3, is also
coding or non-coding RNAs.
of heterochromatin in organisms ranging from yeast to required for silencing and contains a conserved Ago-
plants and mammals10. Additionally, early sequencing binding GW domain23,25,26. The RITS complex associates
experiments detected small RNAs that mapped to the with the RNA-dependent RNA polymerase complex
pericentromeric repeat regions, and Rdp1 was found (RDRC), which includes Rdp1, the helicase Hrr1 and
to associate with centromeric DNA10,11. Based on these the non-canonical poly(A) polymerase Cid12 (REF. 27).
Both complexes associate not only with heterochromatin
and one another, but also with non-coding pericentro-
Box 1 | Small-RNA silencing pathways meric RNAs27. The functional importance of this RNA
association is made clear by the observation that ectopic
RNA interference (RNAi) is used broadly to refer to various RNA silencing pathways tethering of the RITS complex to a euchromatic mRNA
that use small RNAs, together with a member of the conserved Argonaute (AGO) and triggers H3K9 methylation at its site of transcription28.
PIWI family of proteins, to target genes for inactivation at the post-transcriptional or
Together, these results led to the development of the
transcriptional levels1,140. Classical RNAi is triggered by long double-stranded RNA
‘nascent transcript’ model for RNAi-dependent hetero-
(dsRNA) precursors, which are processed to 22–23‑nucleotide duplex small interfering
RNAs (siRNAs) with 2‑nucleotide 3ʹ overhangs by the RNase III ribonuclease chromatin assembly 27,28 (FIG. 1). Chromatin-associated
Dicer141–145. siRNA duplexes contain 5ʹ monophosphate and 3ʹ OH groups, and are RNAs are thought to act as scaffolds for the cooperative
loaded onto AGO family proteins, which associate with the siRNA 5ʹ monophosphate- assembly of complex machineries that couple small-RNA-
containing nucleotide via their middle (MID) domain, and with the siRNA 3ʹ nucleotide mediated recognition to chromatin modifications. This
via their PIWI–AGO–ZWILLE (PAZ) domain146–148. The arrangement and specificity of co-transcriptional gene silencing mechanism contains
the MID and PAZ domains in AGO proteins allow them to associate with small RNAs two primary components28–31. First, the nascent tran-
of specific sizes with distinct termini. siRNAs guide AGO proteins and their associated script is degraded by both RNAi-dependent and RNAi-
complexes to complementary RNAs, which are then targeted for degradation, independent mechanisms, with RNAi-independent
translational repression or transcriptional silencing3–5. In addition to the MID and PAZ
mechanisms involving TRAMP and exosome complexes29.
domains, AGO proteins contain an RNase H‑like domain that can cleave or slice target
Second, RNAi-dependent H3K9 methylation leads to
RNAs, promoting their degradation149. In some organisms, this slicer activity is also
critical for the release of one of the two AGO-bound siRNA strands and the conversion hetero­chromatin formation and transcriptional gene
of duplex siRNA to mature single-stranded siRNA150–153. In Schizosaccharomyces pombe, silencing30,31. This small-RNA-based targeting strategy for
plants, Tetrahymena thermophila and Caenorhabditis elegans, siRNA-programmed heterochromatin assembly is in contrast with that used
AGO recruits an RNA-dependent RNA polymerase (RdRP), which uses the targeted by the distant fungal relative Saccharomyces cerevisiae,
RNA as a template to synthesize a dsRNA substrate for Dicer, thereby amplifying in which specificity is instead determined by site-specific
siRNAs and the RNAi response27,154–156. DNA-binding proteins (reviewed in REF. 32).
The genomes of Drosophila melanogaster and mammals do not seem to encode In addition to the RDRC and the RITS complex, the
RdRPs. However, these organisms harbour another class of small RNAs, nascent transcript heterochromatin assembly platform
PIWI-interacting RNAs (piRNAs), which mediate RNA degradation in the cytoplasm and
also includes the Clr4–Rik1–Cul4 (CLRC) complex, the
DNA or histone methylation in the nucleus157–160. Metazoan piRNAs originate from
Clr4 subunit of which is the sole H3K9 methyltrans-
single-stranded RNA precursors, and their amplification in the D. melanogaster and
mammalian germ lines involves the ‘ping-pong’ cycle whereby piRNA-guided cleavage ferase in S. pombe33–37. The CLRC complex subunit Rik1
of complementary RNAs by one PIWI paralogue generates the 5ʹ ends of new piRNAs associates with both the RDRC and the RITS complex,
that are loaded onto another paralogue, and vice versa, in a process that degrades and the peripheral CLRC member Stc1 also provides
transposon mRNAs84,161,162. It is proposed that piRNA pools are initially derived from a link to the RNAi machinery via its interaction with
random sampling of the transcriptome and then selectively enriched by ping-pong Ago1, although it seems to be dispensable for the inter-
amplification for sequences corresponding to transposons that are actively transcribed action of the CLRC complex with the RITS complex
and that are able to contribute substrates to the cycle85. This idea is reminiscent of subunit Tas3 (REFS 38−40). These physical connections
primal RNAs (priRNAs) in S. pombe. priRNAs also result from cellular RNA sampling by suggest the existence of a feedback loop in which the
Ago1, but they only direct small-RNA amplification at loci where antisense RNA targets
activities of the RITS complex, the RDRC and the CLRC
are available163. This leads to repeat-specific RNAi-dependent siRNA accumulation163.
complex reinforce each other (FIG. 1) (see below).

72 | FEBRUARY 2015 | VOLUME 16 www.nature.com/reviews/genetics

© 2015 Macmillan Publishers Limited. All rights reserved


REVIEWS

Box 2 | Heterochromatic small RNAs


The discovery of small RNAs derived from the
pericentromeric repeats of the fission yeast
Schizosaccharomyces pombe marked the first example of Pol II
heterochromatic small RNAs in any organism11, and cen repeat
studies of S. pombe have remained useful for uncovering the Pol II
modes of biogenesis of this small-RNA class. Non-coding
transcripts generated from pericentromeric DNA repeats
(cen) act as the precursors for heterochromatic small
interfering RNAs (siRNAs) in S. pombe. Initially, it was
hypothesized that heterochromatic small RNAs arise from
Dicer 1 (Dcr1) cleavage of double-stranded RNA (dsRNA) dsRNA priRNA
formed by the base-pairing of complementary
Dcr1
pericentromeric transcripts. According to this hypothesis,
the RNA-dependent RNA polymerase Rdp1 amplifies the
siRNA pool by generating additional Dcr1 substrates10. Primary siRNA
In this scenario, heterochromatic small RNA levels are
expected to be higher in cells lacking Rdp1, in which
pericentromeric transcripts from opposite strands can still
hybridize and undergo Dcr1 processing, than in cells lacking
Dcr1. Initially, no differences in small RNA levels in rdp1− Ago1 Ago1
and dcr1− cells were detected on northern blots or in early
small-RNA sequencing experiments27,163. However, recent
advances in increasing the depth of sequencing have enabled Triman
the detection of a small population of Dcr1‑dependent,
Rdp1‑independent heterochromatic small RNAs known as Rdp1
primary siRNAs, which are thought to arise from the Ago1
base-pairing and Dcr1 processing of bidirectionally
Heterochromatic gene synthesized transcripts164. Dcr1 cleavage
silencing Dcr1 can limit the biogenesis of primary siRNAs, as and siRNA amplification
Silencing of gene expression overexpression of Dcr1 not only increases primary siRNA
within heterochromatin. It was levels but also induces siRNA production from convergently H3K9 methylation
originally thought to exclusively transcribed loci genome-wide164. Thus, limited Dcr1
involve transcriptional gene
availability is likely to represent an adaptation that restricts cen repeat
silencing mechanisms, but
recent findings indicate that
heterochromatic small-RNA biogenesis to a subset of loci
co‑transcriptional degradation with high levels of bidirectional transcription. The low
of nascent RNA, or abundance of Dcr1 and primary siRNAs is compensated for by the Rdp1‑mediated generation of secondary
heterochromatic siRNAs and also by the cooperation of the two enzymes — Dcr1 associates with Rdp1 Nature
andReviews | Genetics
stimulates its
co‑transcriptional gene
silencing, also play important dsRNA synthesis activity50. Consistent with an adaptive relationship between low Dcr1 abundance and Rdp1 activity,
parts in silencing. overexpression of Dcr1 suppresses, to a large extent, the requirement for Rdp1 and for Dsh1, a factor required for Dcr1–
Rdp1 association, in pericentromeric silencing and histone H3 lysine 9 (H3K9) methylation164,165. These observations also
Pericentromeric DNA repeat rule out a critical role for Rdp1 in the direct recruitment of H3K9 methylation activity to chromatin as previously proposed10.
A repeated DNA sequence that
Finally, a class of Dcr1‑independent heterochromatic small RNAs known as primal RNAs (priRNAs), which are capable of
surrounds the centromeres of
most eukaryotic chromosomes.
directing H3K9 methylation in an Ago1‑dependent manner, has been discovered (see the figure)163. This finding, together
These repeats are assembled with the detection of primary siRNAs164, suggests that self-reinforcing feedback loops involving small RNAs and chromatin
into heterochromatin, which modification (BOX 3) may be nucleated by their small RNA components. In support of this idea, priRNAs and siRNAs, which
has been demonstrated are both trimmed to their mature length by the 3ʹ exonuclease Triman, are required for the maintenance of
to have roles in cohesin RNAi-dependent facultative heterochromatin islands and de novo establishment of constitutive pericentromeric
recruitment in fission yeast heterochromatin (see the figure)166.
and mammals, and de novo
centromere assembly in fission Pol II, RNA polymerase II.
yeast.

RNA-induced transcriptional
silencing
The central role of nascent RNAs in the recruit- Raf2–Rik1 complex seems to be distinct from the CLRC
(RITS). A protein complex ment of RNAi and histone-modifying activities has complex, and direct evidence of cooperation between
first identified in prompted investigations of the interactions between the RNAi and DNA replication machineries is lacking.
Schizosaccharomyces pombe. RNA-mediated silencing and other nuclear processes in It has also been proposed that the splicing machinery
In addition to Argonaute 1, the
S. pombe. The CLRC complex components Raf2 (also contributes to heterochromatic silencing by directly
RITS complex contains a GW
domain protein, Tas3, and a known as Cmc2) and Rik1 were recently found to asso- interacting with the RDRC and the nascent transcript
chromodomain protein, Chp1, ciate with Cdc20 (the catalytic subunit of the leading- platform to promote siRNA biogenesis43,44. However,
which tether the complex strand DNA polymerase‑ε) and Mms19 (a conserved earlier biochemical purifications demonstrated that
to the chromosome via regulator of the general transcription factor TFIIH41), the RDRC subunit Cid12 forms RDRC-independent
interactions with nascent
long non-coding RNAs and
suggesting the coordination of DNA replication with complexes with splicing factors27. Moreover, the RNAi
nucleosomes with methylated RNAi-dependent release of RNA polymerase II (Pol II) defect found in several splicing mutants can be par-
histone H3 lysine 9. in the inheritance of heterochromatin42. However, this tially rescued by the introduction of cDNAs that encode

NATURE REVIEWS | GENETICS VOLUME 16 | FEBRUARY 2015 | 73

© 2015 Macmillan Publishers Limited. All rights reserved


REVIEWS

siRNA ARC Indeed, this paradigm of self-reinforcing loops has


found consistent support from experimental findings in
Tas3 Ago1 the small-RNA-dependent epigenetic pathways of both
RITS fungi and plants.
Chp1
Tas3 Ago1 In S. pombe, the discovery that pericentromeric
Chp1 H3K9 methylation is dependent on the RNAi machinery
Dcr1 was quickly followed by the realization that, conversely,
pericentromeric siRNA accumulation also requires the
RDRC TRAMP and H3K9 methyltransferase Clr4 (REFS 27,49). Several physi-
exosome
Ers1 Tas3 Ago1 cal interactions are now known to underlie this mutual
Swi6 Chp1 dependence. First, the bivalent nature of the RITS com-
H3K9me2 or plex — its ability to interact with nascent RNA via its
H3K9me3
siRNA-bound Ago1 subunit and with chromatin via
Centromeric
repeats Pol II its Chp1 subunit — forms the physical basis for a self-
reinforcing positive feedback loop (FIG. 1). Methylated
Pol II
Chp1 Chp2 Swi6 H3K9 stabilizes RITS complex binding via Chp1 (REF. 49),
Ago1 Tas3 Stc1 promoting siRNA amplification by the RITS complex-
RITS Raf1 Rik1 Clr4 SHREC interacting factors RDRC and Dcr1 (REFS 27,50). Second,
Raf2 Cul4 CLRC methylated H3K9 enhances siRNA accumulation by
helping to recruit the RDRC via the heterochromatin
lncRNA
protein 1 (HP1) homologue Swi6 in an interaction
that is directly bridged by Ers1 (REFS 30,51,52). Finally,
direct associations between the siRNA-programmed
Figure 1 |  The ‘nascent transcript’ model and a self-reinforcing RITS complex and the Rik1 and Stc1 subunits of the
Natureepigenetic loop in
Reviews | Genetics
S. pombe. In Schizosaccharomyces pombe, the RNA-induced transcriptional silencing CLRC38–40 ensure that siRNAs in turn feed back on H3K9
(RITS) complex establishes a physical connection between small interfering RNAs methylation states (FIG. 1).
(siRNAs) and heterochromatin by targeting a nascent transcript, and forms the basis of The notion of self-reinforcing epigenetic states that
a self-sustaining feedback mechanism that couples siRNA production to chromatin involve multiple types of signals was first proposed by
modification. A siRNA-targeted centromeric long non-coding (lncRNA) bound to the Eric Selker while working with the filamentous fungus
RITS complex becomes a template for double-stranded RNA (dsRNA) synthesis by Neurospora crassa46. Initial results revealed that DNA
the RNA-dependent RNA polymerase complex (RDRC, which is composed of Rdp1, methylation patterns were dependent on histone
Hrr1 and Cid12) and generation by Dicer 1 (Dcr1) of new siRNAs, leading to further deacetylation, but recent findings suggest that feedback
targeting of the RITS complex after passage of Argonaute (Ago) through the ARC (Ago
loops that involve small RNAs also exist in N. crassa.
siRNA chaperone) complex. The Chp1 subunit of the RITS complex anchors the
complex onto nucleosomes with histone H3 lysine 9 (H3K9) methylation, and The canonical DNA methylation pathway is RNAi-
the RITS complex recruits the Clr4–Rik1–Cul4 (CLRC, of which Clr4 is the independent 53 and acts downstream of, and without
methyltransferase) complex via Rik1 and Stc1 to promote the further spread of feeding back on, H3K9 methylation states54. By con-
H3K9 methylation. The heterochromatin protein 1 (HP1) homologue Swi6 binds to trast, a newly uncovered DNA methylation phenom-
methylated H3K9 and promotes RDRC recruitment and siRNA biogenesis via the enon exhibits self-reinforcing properties and correlates
silencing factor Ers1. Swi6, and particularly the other HP1 protein Chp2, help to with the biogenesis of small RNAs. This type of DNA
restrict RNA polymerase II (Pol II) access by recruiting the Snf2–histone deacetylase methylation occurs at gene-rich and intergenic loci that
repressor complex (SHREC ). The TRAMP non-canonical poly(A) polymerase and the are transcribed in a convergent manner and that also
exosome also contribute to silencing. Together, the RITS complex and the nascent give rise to Dicer-independent siRNAs (disiRNAs)55,56.
lncRNA transcript provide a hub for the assembly of machineries that make siRNAs,
Remarkably, disiRNA loci DNA methylation (DLDM),
modify histones and silence gene expression. me2, dimethylation.
unlike the well-studied DNA methylation pathway asso-
ciated with repetitive sequences, is required for H3K9
trimethylation (H3K9me3) and is thus self-reinforcing
RNAi components45, suggesting that the effects of splic- through the known HP1‑dependent mechanism that
ing mutations on heterochromatic silencing are at least recruits the DNA methyltransferase DIM‑2 to the
partially due to improper splicing of pre-mRNAs that regions of H3K9 methylation54,55. It was also observed
encode RNAi proteins. that, for a given DLDM locus, most clones are not meth-
ylated; however, when methylation does occur clones are
siRNAs form self-reinforcing epigenetic loops with DNA methylated extensively across the disiRNA-producing
and histone methylation. The observation that silent locus, suggesting that a powerful positive feedback loop
states of gene expression and their associated chromatin leads to the spread of the epigenetic mark55. Convergent
modifications can spread along DNA in cis and persist transcription is critical and, when induced artificially, is
throughout cell division32 led to the idea that positive sufficient to drive this dynamic DNA methylation pat-
feedback loops might form the molecular basis for epige- tern55. The potential roles of the disiRNAs themselves,
netic memory. Linking the recognition of signals — such and other mechanistic aspects of DLDM, remain to
as covalent histone modifications, DNA methylation or be elucidated, but these results further underscore the
small RNAs — to the generation of new signals would prevalence of self-reinforcing epigenetic loops among
ensure the stable propagation of silent states46–48 (BOX 3). small-RNA silencing pathways in the nucleus.

74 | FEBRUARY 2015 | VOLUME 16 www.nature.com/reviews/genetics

© 2015 Macmillan Publishers Limited. All rights reserved


REVIEWS

RNA-directed DNA methylation in A. thaliana. In as SUVH4), SUVH5 and SUVH6, contribute to the
A. thaliana, 24‑nucleotide siRNAs direct de novo DNA maintenance of DNA methylation patterns, with KYP
methylation and maintenance of DNA methylation having the primary role58. KYP is recruited to sites
at asymmetrical CHH sites (where H represents any of DNA methylation via its SET and RING finger-
base except G) in a manner that depends on two plant- associated (SRA) domain67 and, conversely, the Pol
specific polymerase II (Pol II)‑related RNA polymerases, IV‑associated factor SHH1 binds to nucleosomes
Pol IV and Pol V57 (reviewed in REF. 58). This phenom- with methylated H3K9 and unmethylated H3K4
enon exhibits many physical connections between via its SAWADEE domain, thus promoting siRNA-
pathway components that form the basis of a self- dependent methylation of the underlying DNA68,69.
reinforcing epigenetic loop (FIG. 2). First, 24‑nucleotide Another example of self-reinforcement involves
siRNAs are generated by DCL3 processing of double- the SUVH2 and SUVH9 proteins which, like KYP,
stranded RNA (dsRNA) synthesized by the RdRP bind to methylated DNA via their SRA domains
RDR2 using Pol IV transcripts as templates. Recent but, unlike KYP, lack H3K9 methyltransferase activ-
in vitro experiments have shown that RDR2 and Pol IV ity 70. Two recent studies indicate that these SRA
associate in an RNA-independent manner and that the domain-containing proteins directly recruit Pol V to
activity of RDR2, but not of Pol IV, requires this associa- sites of pre-existing DNA methylation, thus generat-
tion, which suggests an evolved mechanism for limiting ing nascent transcript scaffolds for recruitment of
siRNA biogenesis to loci occupied by Pol IV59. Loading AGO4–siRNA complexes and further enhancement
of siRNAs onto AGO4 in the cytoplasm triggers its of the DNA methylation signal71,72. This dense network of
import into the nucleus60, where it associates directly positive feedback in A. thaliana suggests that the broad
with siRNA-complementary Pol V transcripts61,62. This evolutionary relevance of self-reinforcing siRNA-driven
sequence-specific recruitment is reinforced by direct epigenetic loops extends beyond the S. pombe system in
interactions between AGO4 and the GW domains of which they were first suggested.
both Pol V and the Pol V transcript-binding protein
KTF1 (also known as SPT5L)25,63. While these multiple Conservation and divergence of nuclear RNAi
associations stabilize the localization of AGO4 along a Below, we discuss examples of small-RNA-mediated
Pol V nascent transcript, the RDM1 protein provides silencing in the animal germ line, including C. elegans,
a physical link between AGO4 and the DRM2 DNA D.  melanogaster and mammals. Nuclear RNAi in
methyltransferase, illustrating how the instruction these systems differs from the examples in S. pombe
to silence gene expression is conveyed from small or A. thaliana in one major aspect. Although animal
RNAs to covalent DNA modifications64. RDM1 also small RNAs act as signals that trigger the modifica-
has an affinity for methylated DNA, which raises the tion of chromatin or DNA, it remains unclear whether
possibility that it can direct the Pol V–KTF1–DRM2– small-RNA amplification involves a self-reinforcing
AGO4 complex to pre-existing sites of DNA methyla-
tion. A recent study shows that AGO4 co‑purifies with
DRM2 in vivo, and that DRM2 preferentially methyl-
ates the strand of DNA that acts as a template for Pol V Box 3 | Self-reinforcing positive feedback loops
transcripts65. The authors propose that as AGO4 binds Self-reinforcing positive feedback loops are formed
to a siRNA-complementary nascent Pol V transcript, it by the functional coupling of different types of signal
directs DRM2 specifically to the template DNA strand as generation events. Studies in Schizosaccharomyces
it emerges from the Pol V exit channel. This study pro- pombe, Arabidopsis thaliana and mammals have
vides an unprecedented level of mechanistic detail on the uncovered the molecular basis of these coupling
feedback between small RNAs and DNA methylation. events by identifying the proteins that recognize
Interestingly, at a subset of loci, deletion of DNA either histone H3 lysine 9 dimethylation (H3K9me2)
methylation by mutations in the histone deacetylase and H3K9me3, or DNA cytosine methylation
HDA6 or the maintenance cytosine methyltransferase (DNA‑5mC), and that recruit enzymes which catalyse
the other methylation events. In some organisms (for
MET1 results in loss of silent locus identity 66. This obser-
example, S. pombe and A. thaliana), the methylation
vation indicates that, at these loci, silent locus identity events are also physically coupled to proteins that
is maintained by self-reinforcing interactions involv- recruit small interfering RNA (siRNA) amplification
ing histone deacetylation and DNA methylation, while loops to chromatin, thus forming self-reinforcing loops
siRNA amplification and silencing rely on cooperation in which the histone or DNA methylation event
between the Pol V–KTF1–DRM2–AGO4 complex and promotes siRNA generation and siRNAs in
Pol IV machineries (see above) and on pre-existing turn promote histone or DNA methylation.
chromatin modifications66. Self-reinforcing loops are thought to help to ensure
Embedded in the RNA-dependent DNA methylation the maintenance of epigenetic information47,48.
pathway are important self-reinforcing feedback mecha-
nisms that have been discovered only lately. For instance,
regions of RNA-directed DNA methylation in A. thaliana H3K9 me2/3 siRNA H3K9 5mC
are also enriched in H3K9 methylation58, and these two
epigenetic signals mutually support one another. Three
distinct H3K9 methyltransferases, KYP (also known
Nature Reviews | Genetics

NATURE REVIEWS | GENETICS VOLUME 16 | FEBRUARY 2015 | 75

© 2015 Macmillan Publishers Limited. All rights reserved


REVIEWS

relationship with the downstream chromatin or DNA post-transcriptional gene silencing of transposons by
modifications. Recent studies of the D. melanogaster contributing to ‘ping-pong’ amplification with its cyto-
Piwi-interacting RNAs (piRNAs), so called because plasmic counterparts Aubergine (Aub) and Ago3, also
they partner with Ago proteins of the largely germline- targets transposons at the transcriptional level76–80. In
specific Piwi subfamily, point in this direction, but ovarian somatic cells, which also harbour piRNAs, most
mammalian pathways seem to lack this characteristic. of the euchromatic H3K9 methylation islands in the
Nuclear RNAi-related events associated with the intro- genome correspond to transposon insertion sites and
duction of foreign siRNAs into animal somatic cells are are Piwi-dependent 79. Furthermore, artificial recruit-
not discussed here. ment of Piwi to a reporter locus induces H3K9 meth-
ylation, HP1a (also known as Su(var)205) accumulation
Small RNAs and chromatin in animal cells: the piRNA and exclusion of RNA Pol II, suggesting that piRNAs
system. piRNAs were first discovered in mice and may have a direct role in guiding chromatin changes76,77.
D. melanogaster, and their roles in silencing transpo- Piwi and HP1a were previously reported to interact
sons in the germ line of these organisms are well estab- directly in vivo 81, but the reproducibility and physio-
lished (reviewed in REF. 73). The C. elegans genome logical importance of this result have been debated77,80,
also encodes piRNAs, called 21U‑RNAs, but they have raising the possibility that there are unidentified links
only been implicated in the repression of one transpos- that transmit instructions from piRNAs to chromatin.
able element 74,75; we discuss the roles of 21U-RNAs in Mechanisms involving Piwi engagement with nascent
epigenetic silencing in the next subsection. transcripts or even the underlying DNA have been
In D. melanogaster, recent evidence suggests that proposed76,77, but understanding the molecular details
the nuclear protein Piwi, in addition to mediating remains a major challenge for future research.

24-nt siRNA

DCL3

Acetylated nucleosome dsRNA Cytoplasmic loading followed by nuclear


import of AGO4–siRNA complex

RDR2
H3K9 methylated
nucleosome

GW
AGO4
DRM2
CMT3 RDM1 KTF1
GW

MET1
H3K9 methylated and
H3K4 unmethylated Pol IV Pol V
nucleosome
SHH1 SWD KYP SRA SUVH2
HDA6 and
Methylated DNA SUVH5 SUVH6 SUVH9

Figure 2 |  A self-reinforcing loop linking siRNAs to DNA and histone methylation, which is deposited by the enzymes KYP, SUVH5 and SUVH6.
methylation in A. thaliana.  Elaborate feedback between small RNAs and Nature
These methylation events are coupled to one another Reviews
and | Genetics
to siRNA activity
DNA and histone methylation underlies a robust silencing pathway at sites in several ways. KYP is recruited directly to methylated DNA, where it
of asymmetrical cytosine methylation in the Arabidopsis thaliana genome. methylates neighbouring histones, and the H3K9 methylation reader SHH1
Two plant-specific polymerases transcribe the critical RNAs. RNA recruits Pol IV to promote siRNA generation, while the DNA methylation
polymerase IV (Pol IV) transcripts are processed by the RNA-dependent readers SUVH2 and SUVH9 recruit Pol V to promote AGO4 targeting and
RNA polymerase RDR2 and the Dicer protein DCL3 into 24‑nucleotide (nt) further DNA methylation. Thus, the different methylation readers, RNA
small interfering (siRNAs), while Pol V transcripts act as their targets. The polymerases and AGO4 act together to create self-reinforcing interactions
Argonaute protein AGO4, the siRNA-dependent recruitment of which to between pre-existing DNA methylation and siRNA amplification. Erasure
Pol V transcripts is reinforced by interactions with the GW domains of Pol V of DNA methylation by mutations in either the histone deacetylase HDA6
and an associated elongation factor KTF1, in turn recruits the CHH DNA or the maintenance DNA methyltransferase MET1 results in loss of siRNA
methyltransferase DRM2. RDM1 associates with the Pol V–AGO4–DRM2 biogenesis, emphasizing the importance of these self-enforcing
complex and may link siRNA amplification to pre-existing DNA interactions. Altogether, the A. thaliana pathway for DNA methylation at
methylation. Meanwhile, another DNA methyltransferase that targets CHG asymmetrical sites is one of the most remarkable examples of a recurring
sites for maintenance, CMT3, is recruited directly to methylated histone theme in epigenetic regulation by small RNAs: self-reinforcing feedback
H3 lysine 9 (H3K9). Silencing by DNA methylation is augmented by H3K9 loops. SRA, SET and RING finger-associated; SWD, SAWADEE domain.

76 | FEBRUARY 2015 | VOLUME 16 www.nature.com/reviews/genetics

© 2015 Macmillan Publishers Limited. All rights reserved


REVIEWS

Recent findings also show that sites bearing Piwi- trans-silencing loci. The new allele also acquired the
dependent H3K9 methylation in turn function as source ability to convert other alleles into trans-silencers in
loci for piRNA production in D. melanogaster ovaries. the same manner; thus, this constitutes a case of
Methylated H3K9 recruits the HP1 family protein Rhino, paramutation90,91. These results indicate that maternally
which, together with the Cutoff protein, promotes non- deposited cytoplasmic piRNAs can direct the conversion
canonical transcription from these loci to generate pre- of homologous sequences into piRNA-generating loci
cursors to be processed into piRNAs82,83. Therefore, not with trans-silencing ability. This work shows how a locus
unlike the plant and fungal systems discussed above, the can become incorporated into the piRNA repertoire and
D. melanogaster piRNA pathway seems to exhibit a posi- how RNA itself can act as a carrier of epigenetic informa-
tive feedback loop in which small RNAs guide histone tion. Future discoveries regarding the molecular basis
methylation, which in turn enables biogenesis of more of this paramutation process will provide fundamental
small RNAs. insights into both piRNA biology and the principles of
In mice, piRNAs silence transposons in the male epigenetic inheritance.
germ line by targeting them for de novo DNA methyl­
ation during late embryonic and early neonatal devel- RNAi and H3K9 methylation in C. elegans. C. elegans
opment 84–87. Consistently, the nuclear PIWI protein, has an array of 27 AGO homologues that associate with
Piwi-like protein 4 (PIWL4; also known as MIWI2), is diverse small RNA populations and that act in distinct
expressed at this time85. Thus, like plants, mammals have silencing steps92. Classical RNAi, induced by exogenous
an RNA-directed DNA methylation pathway, although dsRNA, initially involves the AGO protein RDE‑1 and
few of its mechanistic details are currently understood. primary siRNAs; however, RdRPs generate secondary
Arguing against a self-reinforcing loop is the observation siRNAs that are not loaded onto RDE‑1 (REF. 92). Among
that piRNA levels remain high in DNA (cytosine‑5-)- other roles, secondary siRNAs can trigger RNAi in the
methyltransferase 3‑like (Dnmt3l) mutants, in which nuclei of somatic cells by acting with the AGO protein
transposon DNA methylation is completely compro- NRDE‑3. Upon small-RNA loading, NRDE‑3 enters
mised85. However, it is unclear to what extent H3K9 the nucleus, associates with complementary nascent
methylation at transposons, which could also contrib- transcripts and recruits the non-AGO silencing factor
ute to piRNA generation, is affected in Dnmt3l mutants. NRDE‑2, which in turn promotes H3K9 methylation
One study has suggested that piRNA-directed DNA and inhibits elongation of RNA polymerases beyond
methylation in mice involves a system that is similar to the site targeted by siRNAs17,93,94 (FIG. 3). Thus, as in
the nascent transcript model in S. pombe88. The authors fungi and plants, small RNAs in C. elegans can trigger
found that piRNAs mapped to a long terminal repeat transcriptional gene silencing. However, currently, the
(LTR) adjacent to a paternally imprinted locus, RAS biochemical connections remain obscure except for
protein-specific guanine nucleotide-releasing factor 1 a direct interaction between NRDE‑3 and NRDE‑2
(Rasgrf1), and that piRNA pathway components are (REF. 16). More recent work has shown that additional
required for methylation of Rasgrf1. They also found AGO homologues are also required for chromatin
evidence of sequence-specific piRNA-dependent cleav- targeting 17. Interestingly, RNAi-induced H3K9 methyl­
age of a non-conding RNA overlapping with the LTR, ation is a transgenerational phenomenon, as it is inherited
supporting the nascent transcript theory 88. However, it for at least two generations, in the absence of the
has also been shown that the endonuclease activity of dsRNA trigger, by a mechanism that involves germline
PIWL4 is completely dispensable for DNA methylation transmission of siRNAs17,95.
of long interspersed elements (LINE1) and the intracis- In the C. elegans germ line, inheritance of H3K9
Long terminal repeat ternal A particle (IAP) family of LTR retrotransposons89. methylation induced by exogenous dsRNA is mediated
(LTR). A DNA sequence that This suggests that nascent transcript slicing is not an by the germline-specific AGO protein HRDE‑1 (also
is repeated at the ends of important event in small-RNA-dependent chromatin known as WAGO‑9) in association with RdRP-derived
retrotransposons or pro-viral
silencing in mice, although it has a critical role in piRNA siRNAs96,97. Intriguingly, both HRDE‑1 and NRDE‑2
DNA that is formed from
retroviral RNA by reverse amplification and post-transcriptional gene silencing. are also necessary for germline immortality and fertil-
transcription. Plant and Finally, piRNAs also contribute to epigenetic phenomena ity 97. This biological requirement is reminiscent of the
mammalian genomes contain in which they are themselves responsible for epigenetic piRNA pathway, the widely conserved germline main-
thousands of LTRs. inheritance. This idea first emerged in a landmark tenance system discussed above and represented in
Epigenetic phenomena
paper demonstrating the requirement for maternal C. elegans by the 21U‑RNAs and the PIWI-subfamily
Phenomena in which changes deposition of piRNAs, in addition to the presence of AGO protein PRG‑1 (REF. 74). Indeed, several studies
in gene expression occur genomic piRNA loci, for effective transposon silencing have concluded concurrently that HRDE‑1 is in fact the
without a corresponding and fertility in the offspring 15. A recent study found that downstream effector of piRNA-induced silencing in
change in the DNA sequence;
in D. melanogaster, strains containing certain P‑lacZ the germ line96,98–100. Another observation was also made
such changes are stable in the
absence of initiating signals. transgene insertions, but not other strains containing — while HRDE‑1, NRDE‑2 and the HP1 homologue
similar insertions at the same site, showed silencing HPL‑2 are required to maintain a permanent memory
Paramutation of homologous sequences in trans 90. Remarkably, if of these silencing events at the chromatin level, PRG‑1
The ability of a silent allele to maternally inherited, the trans-silencing-competent and 21U‑RNAs are only required for its initiation96,99,100
convert an active allele to the
silent (and paramutagenic)
alleles could convert the paternally inherited alleles (FIG. 3). Thus, different components are responsible for
form. It was first described in into strong trans-silencers, which was accompanied by the establishment and maintenance steps in this example
Zea mays. de novo production of piRNAs corresponding to these of RNA-mediated epigenetic regulation.

NATURE REVIEWS | GENETICS VOLUME 16 | FEBRUARY 2015 | 77

© 2015 Macmillan Publishers Limited. All rights reserved


REVIEWS

a Somatic nuclear RNAi pathway b Germline non-self and self pathways


Non-self RNA Self RNA
Dicer and PRG-1
loading factors U

RDE-1
RdRP
G
G
G G
Post-transcriptional 22G-RNA
gene silencing G

RdRP RDE-1
HRDE-1 CSR-1
AAA G G

NRDE-3
G
Cytoplasm
Nucleus

NRDE-3 HRDE-1 –

NRDE-2 – NRDE-2 CSR-1


HMTase HPL-2 HMTase HPL-2

Pol II Pol II
H3K9me3

Figure 3 |  Small-RNA-driven transcriptional silencing of gene expression in C. elegans.  a | In Caenorhabditis


Nature Reviews | Genetics
elegans, exogenous double-stranded RNA (dsRNA) is processed into primary small interfering RNAs (siRNAs) that
are loaded onto the Argonaute (AGO) protein RDE‑1 and amplified by RNA-dependent RNA polymerases (RdRPs)
to give rise to secondary siRNAs called 22G‑RNAs. When loaded with 22G‑RNAs, the somatic AGO protein NRDE‑3
translocates to the nucleus, where it targets nascent RNA transcripts and silences corresponding genes, acting in
concert with the silencing factor NRDE‑2. Gene expression is halted by NRDE‑2 during the elongation phase of
transcription, and silencing involves histone H3 lysine 9 trimethylation (H3K9me3) and recruitment of the
heterochromatin protein 1 (HP1)‑like protein HPL‑2. b | In the germ line, small-RNA-directed transcriptional
silencing is mediated not by NRDE‑3 but by a different AGO protein, HRDE‑1, which also acts through NRDE‑2,
H3K9 methylation and HPL‑2. HRDE‑1 receives 22G‑RNA inputs both from the pathway that responds to
exogenous dsRNA and from the PIWI-interacting RNA (piRNA), or 21U‑RNA, pathway that scans the transcriptome
for foreign RNAs. 21U‑RNA-programmed PRG‑1 promotes the RdRP-dependent generation of 22G‑RNAs, which
are loaded onto HRDE‑1. In both cases, HRDE‑1 maintains a persistent, transgenerational memory of silenced
genes in the germ line. Meanwhile, another AGO protein called CSR‑1 binds to 22G‑RNAs that represent the full
complement of endogenously expressed RNAs, and protects the corresponding loci from possible silencing by
HRDE‑1. Thus, the 22G‑RNAs bound by CSR‑1 and HRDE‑1 transmit a germline memory of ‘self’ and ‘non-self’
RNAs, respectively, to be appropriately licensed for expression or silenced. HMTase, histone methyltransferase;
Pol II, RNA polymerase II.

In contrast to piRNAs in D. melanogaster and mam- is the AGO protein CSR‑1, the RdRP-dependent small
mals, C. elegans piRNAs only silence one transposon RNA partners of which map not to silenced loci but
family 74,75. However, transgene insertions can be silenced instead to all mRNAs expressed in the germ line101.
in a PRG‑1‑dependent manner 96,99,100, supporting a con- This was confirmed by recent experiments in which
served function for piRNAs in recognizing ‘non-self ’ CSR‑1‑dependent activity was shown to counteract
nucleic acids. Interestingly, PRG‑1 was found to initiate PRG‑1‑mediated silencing in a heritable manner 102,103.
epigenetic silencing of the same single-copy transgene Thus, PRG‑1, associated with 21U‑RNAs, is proposed
in some lines but not others100. Genetic crosses indicate to sample cellular transcripts for non-self sequences to
that transgene expression states can be converted in be silenced by HRDE‑1, while CSR‑1 carries a reper-
both directions by trans-acting silencing and activating toire of self-expressed sequences to be protected from
factors, leading to the proposal that HRDE‑1–siRNA silencing 96,99,100,102,103 (FIG. 3). CSR‑1 seems to have a criti-
complexes enforce the maintenance of PRG‑1‑directed cal role in sperm, where it promotes spermiogenic gene
silencing, whereas another system prevents or even expression and where its associated small RNAs reflect
reverses its establishment 100. The suggested candidate previous germline expression patterns104.

78 | FEBRUARY 2015 | VOLUME 16 www.nature.com/reviews/genetics

© 2015 Macmillan Publishers Limited. All rights reserved


REVIEWS

RNA-dependent histone methylation without RNAi scenarios (reviewed in REF. 115). The number of RNAs
In the RNAi-mediated chromatin silencing pathways found to be associated with PRC2 in cells is enor-
discussed above, a recurring feature is the role of RNA mous116–119, highlighting the risk of interpreting indi-
as an assembly scaffold. Below, we focus on analogous vidual interactions as specific. For instance, a recent
but RNAi-independent strategies used by both mRNAs study revealed that the binding affinity of PRC2 for the
and lncRNAs to recruit histone-modifying enzymes. bacterial MBP mRNA is equal to its binding affinity for
LncRNAs, typically defined as transcripts more than HOTAIR, and that RNA length is a far better predictor of
200 nucleotides in length that do not encode proteins, PRC2 binding affinity than RNA sequence116. Thus, the
have been uncovered in large numbers as a result of direct interactions reported between PRC2 and lncRNAs,
rapid advances in high-throughput technologies. The such as HOTAIR, RepA and Kcnq1ot1 (REFS 110–112)
function of lncRNAs in genome regulation, although still among others, offer only minimal insight into the regu-
largely unknown, seems to be diverse and is the sub- lation of methyltransferase recruitment to particular
ject of a growing field of study (reviewed in REF. 105). A genomic loci.
common theme in current models is the ability of many Although the currently prevailing models concerning
lncRNAs to target chromatin-modifying activities to Polycomb recruitment by lncRNAs remain inconclusive,
particular genomic sites. Perhaps the most extensively the reality is likely to involve a much subtler and more
studied examples are those that involve dosage compen- complex system of sequentially and/or cooperatively
sation in metazoans (reviewed in REF. 106), including assembled components, in which specific lncRNAs act
the mammalian lncRNA XIST, which mediates global as scaffolds. For example, the Jumonji family protein
inactivation of a randomly chosen X chromosome in JARID2 was recently reported to act as an essential inter-
females in a process known as X chromosome inacti- mediate between Xist and PRC2 (REF. 114). While the A
vation (XCI). XIST has been proposed to function by repeat of Xist is required for gene silencing, it does not
directing the Polycomb repressive complex 2 (PRC2) seem to be required for the recruitment of PRC2, as pre-
H3K27 methyltransferase to chromatin. Below, we viously proposed. Instead, Xist recruits JARID2 via its B
examine the evidence and unresolved questions regard- and F repeats, and JARID2 is required for the localization
ing the recruitment of Polycomb to chromatin by XIST of PRC2 and methylation of H3K27 (REF. 114). Another
and other lncRNAs. The basis of specificity for the pro- study has suggested that the facilitation of JARID2–
posed protein–RNA interactions remains unknown. PRC2 interactions may represent a more general func-
We also discuss a class of lncRNAs transcribed from tion of lncRNAs120, although the amino acid residues that
enhancers that are thought to activate gene expression were thought to mediate JARID2 RNA binding differ
in cis. Finally, we describe a case in which the mode of between the two studies114,120. Thus, the events linking
specificity is better understood — the targeting of the lncRNAs to the recruitment of PRC2, and other meth-
Clr4 H3K9 methyltransferase by the mRNAs of meiotic yltransferases to chromatin, are likely to be multiple
genes to their gene loci in S. pombe. and intricate, involving the contributions of other pro-
tein factors and perhaps other histone modifications115.
lncRNAs and Polycomb recruitment. Polycomb pro- The increasing scepticism towards widespread assump-
teins and H3K27me3 were first linked to XCI by genetic tions is promising and should ultimately lead to a better
and cytological experiments107–109. A more direct bio- understanding of how lncRNAs influence chromatin.
chemical connection was shown by the detection of
XIST RNA in PRC2 immunoprecipitation experi- lncRNAs and enhancer function. Transcription at a large
ments and a corresponding in vitro test of their inter- number of mammalian enhancers gives rise to lncRNAs,
action110. REPA — an independent internal transcript called enhancer RNAs (eRNAs), that seem to have major
within XIST, so called because it spans the conserved roles in the regulation of transcription. At least a subset
A repeat region of XIST — was proposed to interact of eRNAs seem to act as nascent transcripts that func-
directly with and recruit the PRC2 catalytic subunit tion in cis to promote the transcription of neighbour-
EZH2 in the earliest stages of XCI establishment. The ing genes. Initially, high-throughput RNA sequencing
same region of full-length XIST would then directly experiments identified thousands of eRNAs that are
recruit PRC2 as silencing spreads in cis110. It was also transcribed from enhancer elements in response to sig-
reported that other lncRNAs—such as KCNQ1 opposite nals that mediate enhancer-dependent transcriptional
strand/antisense transcript 1 (Kcnq1ot1) and HOX activation121,122. RNAi-mediated knockdown of several
transcript antisense RNA (HOTAIR)—could directly eRNAs was then shown to result in reduced expression
recruit PRC2 to chromatin in cis or in trans, respec- of proximally located target mRNAs, suggesting that
tively111,112. However, one study failed to detect an inter- eRNAs act in cis and are required for the activation of
action between PRC2 and the XIST A repeats using an target gene transcription123,124.
ultraviolet crosslinking approach113, and others have Insights into the mechanism of action of eRNAs
found the XIST A repeat to be dispensable for PRC2 came from the identification of components of the
chromatin association and H3K27 methylation in mice transcription machinery that mediate eRNA function.
and humans107,114. More generally, the idea that lncRNAs eRNAs involved in the activation of the developmen-
mediate PRC2 targeting through direct interactions has tally regulated genes T-cell acute lymphocytic leukae-
been undermined both by doubts regarding specificity mia 1 (TAL1), snail family zinc-finger 1 (SNAI1) and
and by the existence of equally compelling alternative SNAI22 require the Mediator transcription co‑activator

NATURE REVIEWS | GENETICS VOLUME 16 | FEBRUARY 2015 | 79

© 2015 Macmillan Publishers Limited. All rights reserved


REVIEWS

a activation. Large RNAs (>200 nucleotides) may be par-


DNA-binding Histone or DNA
proteins E
methylation ticularly suited for such architectural tasks that bring
enhancer and promoter regions, usually located great
distances apart, into proximity.
The identification of both the regions within eRNAs
that mediate co‑activator complex recruitment and the
RNA-binding domains in the subunits of these com-
plexes is required for a better understanding of how
different eRNAs activate transcription. In addition to cis-
acting eRNA, several trans-acting lncRNAs that activate
transcription have also been identified. They are not dis-
b cussed here but have recently been extensively reviewed
AGO or PIWI
E elsewhere22,127.
Histone or DNA
methylation
Silencing of meiotic genes in S. pombe. The regulation of
certain meiotic genes in S. pombe is another well-studied
Pol II example in which an RNA scaffold promotes the RNAi-
independent recruitment of a histone methyltransferase
to chromatin. Meiosis-specific genes are silenced in veg-
etative cells by a post-transcriptional process that tar-
gets an RNA cis-element called determinant of selective
c removal (DSR)128. Remarkably, DSRs are found within the
RNA-binding
protein protein-coding regions of meiotic mRNAs128. Therefore,
E they must exhibit some degree of sequence flexibility to
Histone or DNA preserve meiotic protein structure but effectively consist
methylation ? of different variants of a hexanucleotide motif 129. At a
subset of DSR-containing genes, notably mei4+ and ssm4+,
Pol II H3K9 methylation is also observed and depends on the
DSR and transcription, which suggests that the mRNAs
of these genes direct Clr4 to chromatin and may promote
Figure 4 |  RNAs, both short and long, represent an alternative to DNA-binding transcriptional and post-transcriptional silencing 130,131.
Nature Reviews | Genetics
proteins as specificity determinants for epigenetic regulation of gene expression.  How do DSR-containing mRNAs recruit the Clr4
Enzymes (E) that catalyse methylation of histone tails or cytosine bases in DNA are methyltransferase? Much like lncRNAs and Polycomb
recruited to chromatin by distinct mechanisms. a | Sequence-specific DNA-binding
proteins, the process seems to involve a host of factors
proteins recruit histone- or DNA-modifying enzymes to chromatin. b | Small RNAs
target an Argonaute (AGO) or PIWI protein to a nascent transcript through rather than direct interactions between the RNA and
base-pairing interactions to recruit modifying enzymes. c | Long RNAs act as scaffolds the histone-modifying enzyme. In the case of meiotic
for RNA-binding proteins to recruit chromatin-modifying complexes. In all cases, the mRNAs, the source of specific recognition is much
binding of the enzyme or the recruiting factors (for example, AGO–PIWI complexes in clearer. DSRs are bound directly by the YTH-domain
part b and RNA-binding protein complexes in part c) to chromatin may be enhanced protein Mmi1, and this binding event is required for mei-
by interactions with pre-existing modifications, which self-reinforce the epigenetic otic gene silencing and for H3K9 methylation of the mei4+
state. Pol II, RNA polymerase II. and ssm4+ loci128,129,131. The Mmi1‑interacting zinc-finger
protein Red1 is also required for silencing and H3K9
methylation at mei4+ and ssm4+ (REFS131,132). Strikingly,
for their activity 125, while HOXA distal transcript anti- Red1 associates with Clr4, which suggests that it serves as
sense RNA (HOTTIP), an eRNA involved in the activa- the critical link that targets histone methylation to genes
tion of HOXA homeobox genes, functions through the producing DSR-bearing transcripts130,131. Interestingly, this
WDR5–MLL (also known as KMT2A) H3K4 methyl- Mmi1–Red1–Clr4 axis of RNA-directed H3K9 methyl­
transferase complex 124. Chromosome conformation cap- ation also seems to operate outside the meiotic process
ture experiments show that the eRNA mediates looping to target context-specific genes such as pho1+, which is
interactions between the enhancer and promoter regions repressed when phosphate is available. Data from two
of genes. These studies have given rise to an attractive recent studies show that an upstream lncRNA mediates
model in which eRNAs recruit co‑activator complexes Red1 chromatin binding and H3K9 methylation at the
and promote their interaction with gene promoters to pho1+ locus in a phosphate- and Mmi1‑dependent man-
activate transcription. Consistent with this model, there ner 133,134. Importantly, the lncRNA was found to contain
is evidence that supports a role for specific enhancer DSR motifs that are critical for this regulation134.
sequences that encode eRNAs, which are different from While the DSR-bearing mRNAs demonstrate RNA-
enhancer sequences that form binding sites for transcrip- mediated recruitment of histone modifications with-
tion factors, in activation of target genes126. These studies out RNAi, silencing itself seems to be independent of
provide a new function for nascent RNA transcripts as H3K9 methylation. This is perhaps made most evident
scaffolds for the recruitment of co‑activator complexes by the observation that some DSR-containing genes are
that mediate chromosome looping and transcriptional silenced by Mmi1 and Red1 without exhibiting H3K9

80 | FEBRUARY 2015 | VOLUME 16 www.nature.com/reviews/genetics

© 2015 Macmillan Publishers Limited. All rights reserved


REVIEWS

methylation 131. DSRs inserted ectopically in other (FIG. 4a). The second major principle involves the role of
genes produce similar effects 130. Most conclusively, RNA as a component of self-reinforcing positive feed-
cells lacking Clr4 show wild-type silencing of all tested back loops. These loops are unique to small-RNA sys-
meiotic genes during vegetative growth 135. Instead, tems that contain an amplification component and have
the critical event for silencing of these genes seems to key roles in the epigenetic inheritance of histone and
involve degradation by the nuclear exosome through DNA methylation patterns. The key event is the locali-
a mechanism that requires polyadenylation of the tar- zation of the small-RNA amplification machineries
get RNAs136. Consistently, Red1 interacts with the exo- on nascent transcripts, and their activation by the his-
some subunit Rrp6 (REF. 132), and proteomic analysis tone or DNA methylation events induced by the small
recently showed that this association is mediated by the RNAs themselves (BOX 2). Thus, in both S. pombe and
Mtr4‑like helicase Mtl1, another recently identified Red1 A. thaliana, small-RNA amplification and histone or
interactor 133,135. This poses the question of what fitness DNA methylation are co‑dependent. Self-reinforcing
advantage is conferred by H3K9 methylation at meiotic loops have not yet been described for lncRNAs that act
genes, if it is not necessary for silencing. The presence independently of RNAi. However, cooperative recruit-
of H3K9 methylation might reflect an ancestral path- ment involving the association of chromatin-modifying
way that predates the contribution of the exosome and complexes with both pre-existing chromatin modifica-
that has not yet been lost 135. It is also possible that H3K9 tions and DNA-binding specificity factors has been sug-
methylation could control expression of DSR-containing gested to contribute to the maintenance of epigenetic
genes under certain circumstances, for example, during states in budding yeast 32. By analogy, it is possible that
gene induction134,135. Additional work is needed in this direct or indirect interactions between RNA-binding
area not only to define the biological function of histone proteins that recognize lncRNA and also pre-existing
methylation but also to further isolate the steps that lead histone modifications help to reinforce lncRNA-medi-
to Clr4 recruitment, for example, to determine whether ated changes in chromatin structure. Finally, small
Mmi1 and Red1 accomplish this task independently. RNAs, and their association with positive feedback loops
Studies of the meiotic mRNAs in S. pombe have pro- in the germ line, allow them to act as components of
vided a valuable paradigm for investigating the RNA- transgenerational inheritance mechanisms. The trans-
mediated targeting of histone methyltransferases in all mission of small RNAs through meiosis seems to act as a
its mechanistic complexity. In this regard, it is worth signal for the inheritance of internal or environmentally
noting that the XIST RNA is thought to have evolved induced changes from parent to offspring 138,139.
from a protein-coding gene137. This XIST precursor Despite the considerable progress outlined in this
gene may have shared certain features of the S. pombe Review, several important questions about the biogenesis
meiotic RNAs that enabled it to locally recruit repressive and function of non-coding RNAs remain unanswered.
chromatin-modifying complexes before it acquired the The mechanisms that distinguish between different
ability to spread along the entire X chromosome. types of transcription and that trigger the generation of
different classes of small RNAs remain to be fully under-
Conclusions and perspectives stood, although the available evidence indicates a major
The mechanisms by which coding RNAs and non- role for RNA processing events that act co‑transcription-
coding RNAs regulate chromatin structure in different ally to determine whether a nascent transcript becomes a
organisms share key similarities that allow us to note functional mRNA or is marked for processing by RNAi
common principles which unify these ancient path- and other surveillance mechanisms. Finally, the mecha-
ways (FIG. 4). The first is the principle of recruitment via nisms by which lncRNAs participate in the recruitment
nascent RNA. In these systems, recruitment of effector of Polycomb proteins and other chromatin-modifying
complexes that methylate histones or DNA involves the activities, particularly the molecular basis of specificity,
association of small-RNA-guided AGO complexes or remain poorly defined. We can look forward to answers
site-specific RNA-binding proteins with nascent coding to these questions and, if the recent past is a guide, to
RNA or non-coding RNA scaffolds (FIG. 4b,c), rather more exciting and unexpected discoveries about the
than specific sites on DNA and DNA-binding proteins roles of RNA in gene regulation.

1. Moazed, D. Small RNAs in transcriptional gene silencing 7. Dalmay, T., Hamilton, A., Mueller, E. & 12. Matzke, M., Matzke, A. J. & Kooter, J. M. RNA: guiding
and genome defence. Nature 457, 413–420 (2009). Baulcombe, D. C. Potato virus X amplicons in gene silencing. Science 293, 1080–1083 (2001).
2. Cech, T. R. & Steitz, J. A. The noncoding RNA Arabidopsis mediate genetic and epigenetic gene 13. Fire, A. et al. Potent and specific genetic interference
revolution — trashing old rules to forge new ones. Cell silencing. Plant Cell 12, 369–379 (2000). by double-stranded RNA in Caenorhabditis elegans.
157, 77–94 (2014). 8. Jones, L. et al. RNA–DNA interactions and DNA Nature 391, 806–811 (1998).
3. Hammond, S. M., Boettcher, S., Caudy, A. A., methylation in post-transcriptional gene silencing. 14. Mochizuki, K., Fine, N. A., Fujisawa, T. &
Kobayashi, R. & Hannon, G. J. Argonaute2, a link Plant Cell 11, 2291–2301 (1999). Gorovsky, M. A. Analysis of a PIWI-related gene
between genetic and biochemical analyses of RNAi. 9. Mette, M. F., van der Winden, J., Matzke, M. A. & implicates small RNAs in genome rearrangement in
Science 293, 1146–1150 (2001). Matzke, A. J. Production of aberrant promoter tetrahymena. Cell 110, 689–699 (2002).
4. Hutvagner, G. & Zamore, P. D. A microRNA in a transcripts contributes to methylation and silencing of References 6–8, 10, 11 and 14 provided the first
multiple-turnover RNAi enzyme complex. Science unlinked homologous promoters in trans. EMBO J. evidence for the role of small RNAs in chromatin
297, 2056–2060 (2002). 18, 241–248 (1999). modification and transcriptional gene silencing.
5. Meister, G. et al. Human Argonaute2 mediates RNA 10. Volpe, T. A. et al. Regulation of heterochromatic 15. Brennecke, J. et al. An epigenetic role for maternally
cleavage targeted by mi­­RNAs and siRNAs. Mol. Cell silencing and histone H3 lysine‑9 methylation by inherited piRNAs in transposon silencing. Science
15, 185–197 (2004). RNAi. Science 297, 1833–1837 (2002). 322, 1387–1392 (2008).
6. Hamilton, A. J. & Baulcombe, D. C. A species of small 11. Reinhart, B. J. & Bartel, D. P. Small RNAs correspond 16. Guang, S. et al. Small regulatory RNAs inhibit RNA
antisense RNA in posttranscriptional gene silencing in to centromere heterochromatic repeats. Science 297, polymerase II during the elongation phase of
plants. Science 286, 950–952 (1999). 1831 (2002). transcription. Nature 465, 1097–1101 (2010).

NATURE REVIEWS | GENETICS VOLUME 16 | FEBRUARY 2015 | 81

© 2015 Macmillan Publishers Limited. All rights reserved


REVIEWS

17. Gu, S. G. et al. Amplification of siRNA 36. Jia, S., Kobayashi, R. & Grewal, S. I. Ubiquitin ligase 61. Wierzbicki, A. T., Haag, J. R. & Pikaard, C. S.
in Caenorhabditis elegans generates a component Cul4 associates with Clr4 histone Noncoding transcription by RNA polymerase Pol IVb/
transgenerational sequence-targeted histone H3 methyltransferase to assemble heterochromatin. Pol V mediates transcriptional silencing of overlapping
lysine 9 methylation footprint. Nature Genet. 44, Nature Cell Biol. 7, 1007–1013 (2005). and adjacent genes. Cell 135, 635–648 (2008).
157–164 (2012). 37. Li, F. et al. Two novel proteins, dos1 and dos2, 62. Wierzbicki, A. T., Ream, T. S., Haag, J. R. & Pikaard, C. S.
This study provides convincing evidence that interact with rik1 to regulate heterochromatic RNA RNA polymerase V transcription guides ARGONAUTE4
classical RNAi in C. elegans results in specific H3K9 interference and histone modification. Curr. Biol. 15, to chromatin. Nature Genet. 41, 630–634 (2009).
methylation at targeted loci. 1448–1457 (2005). This paper provided the first evidence that nascent
18. Rinn, J. L. & Chang, H. Y. Genome regulation by long 38. Bayne, E. H. et al. Stc1: a critical link between RNAi Pol V transcripts act as scaffolds for the recruitment
noncoding RNAs. Annu. Rev. Biochem. 81, 145–166 and chromatin modification required for of AGO4.
(2012). heterochromatin integrity. Cell 140, 666–677 (2010). 63. He, X. J. et al. NRPD4, a protein related to the RPB4
19. Lee, J. T. & Bartolomei, M. S. X‑inactivation, 39. Gerace, E. L., Halic, M. & Moazed, D. The subunit of RNA polymerase II, is a component of RNA
imprinting, and long noncoding RNAs in health and methyltransferase activity of Clr4Suv39h triggers polymerases IV and V and is required for RNA-directed
disease. Cell 152, 1308–1323 (2013). RNAi independently of histone H3K9 methylation. DNA methylation. Genes Dev. 23, 318–330 (2009).
20. Schulz, E. G. & Heard, E. Role and control of Mol. Cell 39, 360–372 (2010). 64. Gao, Z. et al. An RNA polymerase II- and
X chromosome dosage in mammalian development. 40. Zhang, K., Mosch, K., Fischle, W. & Grewal, S. I. AGO4‑associated protein acts in RNA-directed DNA
Curr. Opin. Genet. Dev. 23, 109–115 (2013). Roles of the Clr4 methyltransferase complex in methylation. Nature 465, 106–109 (2010).
21. Yang, P. K. & Kuroda, M. I. Noncoding RNAs and nucleation, spreading and maintenance of 65. Zhong, X. et al. Molecular mechanism of action of
intranuclear positioning in monoallelic gene heterochromatin. Nature Struct. Mol. Biol. (2008). plant DRM de novo DNA methyltransferases. Cell
expression. Cell 128, 777–786 (2007). 41. Li, F., Martienssen, R. & Cande, W. Z. Coordination of 157, 1050–1060 (2014).
22. Bonasio, R. & Shiekhattar, R. Regulation of DNA replication and histone modification by the 66. Blevins, T. et al. A two-step process for epigenetic
transcription by long noncoding RNAs. Annu. Rev. Rik1–Dos2 complex. Nature 475, 244–248 (2011). inheritance in Arabidopsis. Mol. Cell 54, 30–42 (2014).
Genet. 48, 433–455 (2014). 42. Zaratiegui, M. et al. RNAi promotes heterochromatic This paper shows that histone deacetylation and
23. Verdel, A. et al. RNAi-mediated targeting of silencing through replication-coupled release of RNA DNA methylation mark a locus for epigenetic
heterochromatin by the RITS complex. Science 303, Pol II. Nature 479, 135–138 (2011). inheritance and siRNA-mediated silencing.
672–676 (2004). 43. Bayne, E. H. et al. Splicing factors facilitate 67. Johnson, L. M. et al. The SRA methyl-cytosine-binding
This study identifies a nuclear AGO complex, RITS, RNAi-directed silencing in fission yeast. Science domain links DNA and histone methylation. Curr. Biol.
which uses Dicer-generated small RNAs to target 322, 602–606 (2008). 17, 379–384 (2007).
specific chromosome regions for heterochromatin 44. Chinen, M., Morita, M., Fukumura, K. & Tani, T. 68. Law, J. A. et al. Polymerase IV occupancy at RNA-
formation. The RITS complex provided the first Involvement of the spliceosomal U4 small nuclear directed DNA methylation sites requires SHH1.
direct physical link between RNAi and RNA in heterochromatic gene silencing at fission yeast Nature 498, 385–389 (2013).
heterochromatin. centromeres. J. Biol. Chem. 285, 5630–5638 (2010). This paper shows that SHH1 recognizes H3 tails
24. Partridge, J. F., Scott, K. S., Bannister, A. J., 45. Kallgren, S. P. et al. The proper splicing of RNAi with unmodified lysine 4 and methylated lysine 9
Kouzarides, T. & Allshire, R. C. Cis-acting DNA from factors is critical for pericentric heterochromatin residues, and recruits Pol IV to chromatin to
fission yeast centromeres mediates histone H3 assembly in fission yeast. PLoS Genet. 10, e1004334 promote siRNA generation. This provides a clear
methylation and recruitment of silencing factors and (2014). basis for self-reinforcement between histone
cohesin to an ectopic site. Curr. Biol. 12, 1652–1660 46. Selker, E. U. Trichostatin A causes selective loss of modification and small RNA biogenesis in plants.
(2002). DNA methylation in Neurospora. Proc. Natl Acad. Sci. 69. Zhang, H. et al. DTF1 is a core component of
25. El‑Shami, M. et al. Reiterated WG–GW motifs form USA 95, 9430–9435 (1998). RNA-directed DNA methylation and may assist in the
functionally and evolutionarily conserved Argonaute- This paper was the first to propose self-reinforcing recruitment of Pol IV. Proc. Natl Acad. Sci. USA 110,
binding platforms in RNAi-related components. Genes relationships between different types of chromatin 8290–8295 (2013).
Dev. 21, 2539–2544 (2007). modifications: DNA methylation and histone 70. Johnson, L. M., Law, J. A., Khattar, A.,
26. Till, S. et al. A conserved motif in Argonaute- deacetylation. Henderson, I. R. & Jacobsen, S. E. SRA-domain
interacting proteins mediates functional interactions 47. Selker, E. U. Gene silencing: repeats that count. Cell proteins required for DRM2‑mediated de novo DNA
through the Argonaute PIWI domain. Nature Struct. 97, 157–160 (1999). methylation. PLoS Genet. 4, e1000280 (2008).
Mol. Biol. 14, 897–903 (2007). 48. Richards, E. J. & Elgin, S. C. Epigenetic codes for 71. Johnson, L. M. et al. SRA- and SET-domain-containing
27. Motamedi, M. R. et al. Two RNAi complexes, RITS and heterochromatin formation and silencing: rounding up proteins link RNA polymerase V occupancy to DNA
RDRC, physically interact and localize to noncoding the usual suspects. Cell 108, 489–500 (2002). methylation. Nature 507, 124–128 (2014).
centromeric RNAs. Cell 119, 789–802 (2004). 49. Noma, K. et al. RITS acts in cis to promote RNA This study shows that the putative methyl
This study reports the first usage of RNA interference-mediated transcriptional and post- DNA-binding proteins SUVH2 and SUVH9 recruit
immunoprecipitation experiments for a transcriptional silencing. Nature Genet. 36, Pol V to sites of pre-existing DNA methylation. By
chromatin-associated protein. 1174–1180 (2004). promoting further RNA-directed DNA methylation,
28. Buhler, M., Verdel, A. & Moazed, D. Tethering RITS to 50. Colmenares, S. U., Buker, S. M., Buhler, M., Dlakic, M. SUVH2 and SUVH9 thus act as key members of a
a nascent transcript initiates RNAi- and & Moazed, D. Coupling of double-stranded RNA self-reinforcing loop.
heterochromatin-dependent gene silencing. Cell 125, synthesis and siRNA generation in fission yeast RNAi. 72. Liu, Z. W. et al. The SET domain proteins SUVH2 and
873–886 (2006). Mol. Cell 27, 449–461 (2007). SUVH9 are required for Pol V occupancy at RNA-
29. Buhler, M., Haas, W., Gygi, S. P. & Moazed, D. 51. Hayashi, A. et al. Heterochromatin protein 1 directed DNA methylation loci. PLoS Genet. 10,
RNAi-dependent and -independent RNA turnover homologue Swi6 acts in concert with Ers1 to regulate e1003948 (2014).
mechanisms contribute to heterochromatic gene RNAi-directed heterochromatin assembly. Proc. Natl 73. Malone, C. D. & Hannon, G. J. Small RNAs as
silencing. Cell 129, 707–721 (2007). Acad. Sci. USA 109, 6159–6164 (2012). guardians of the genome. Cell 136, 656–668 (2009).
This was the first study to demonstrate a major 52. Rougemaille, M. et al. Ers1 links HP1 to RNAi. 74. Batista, P. J. et al. PRG‑1 and 21U‑RNAs interact to
role for co‑transcriptional RNAi-, and TRAMP- and Proc. Natl Acad. Sci. USA 109, 11258–11263 (2012). form the piRNA complex required for fertility in
exosome-dependent RNA degradation in 53. Freitag, M. et al. DNA methylation is independent of C. elegans. Mol. Cell 31, 67–78 (2008).
heterochromatic gene silencing. Together with RNA interference in Neurospora. Science 304, 1939 75. Das, P. P. et al. PIWI and piRNAs act upstream of an
references 27 and 28, it provides evidence for (2004). endogenous siRNA pathway to suppress Tc3
nascent non-coding RNAs as scaffolds for the 54. Honda, S. & Selker, E. U. Direct interaction between transposon mobility in the Caenorhabditis elegans
assembly of RITS and other complexes that DNA methyltransferase DIM‑2 and HP1 is required for germline. Mol. Cell 31, 79–90 (2008).
mediate heterochromatin formation. DNA methylation in Neurospora crassa. Mol. Cell. 76. Huang, X. A. et al. A major epigenetic programming
30. Motamedi, M. R. et al. HP1 proteins form distinct Biol. 28, 6044–6055 (2008). mechanism guided by piRNAs. Dev. Cell 24, 502–516
complexes and mediate heterochromatic gene 55. Dang, Y., Li, L., Guo, W., Xue, Z. & Liu, Y. Convergent (2013).
silencing by nonoverlapping mechanisms. Mol. Cell transcription induces dynamic DNA methylation at 77. Le Thomas, A. et al. Piwi induces piRNA-guided
32, 778–790 (2008). disiRNA loci. PLoS Genet. 9, e1003761 (2013). transcriptional silencing and establishment of a
31. Sugiyama, T. et al. SHREC, an effector complex for 56. Lee, H. C. et al. The DNA/RNA-dependent RNA repressive chromatin state. Genes Dev. 27, 390–399
heterochromatic transcriptional silencing. Cell 128, polymerase QDE‑1 generates aberrant RNA and (2013).
491–504 (2007). dsRNA for RNAi in a process requiring replication Together with reference 76, this paper demonstrates
32. Moazed, D. Mechanisms for the inheritance of protein A and a DNA helicase. PLoS Biol. 8, that ectopic targeting of Piwi in D. melanogaster
chromatin states. Cell 146, 510–518 (2011). e1000496 (2010). using artificially generated piRNAs leads to H3K9
33. Nakayama, J., Rice, J. C., Strahl, B. D., Allis, C. D. & 57. Ream, T. S. et al. Subunit compositions of the methylation and transcriptional silencing at target
Grewal, S. I. Role of histone H3 lysine 9 methylation in RNA-silencing enzymes Pol IV and Pol V reveal their loci. This suggests that piRNA-progammed Piwi can
epigenetic control of heterochromatin assembly. origins as specialized forms of RNA polymerase II. interact with nascent transcripts to induce
Science 292, 110–113 (2001). Mol. Cell 33, 192–203 (2009). chromatin changes.
34. Hong, E. J., Villen, J., Gerace, E. L., Gygi, S. P. & 58. Matzke, M. A. & Mosher, R. A. RNA-directed DNA 78. Rozhkov, N. V., Hammell, M. & Hannon, G. J.
Moazed, D. A cullin E3 ubiquitin ligase complex methylation: an epigenetic pathway of increasing Multiple roles for Piwi in silencing Drosophila
associates with Rik1 and the Clr4 histone H3‑K9 complexity. Nature Rev. Genet. 15, 394–408 (2014). transposons. Genes Dev. 27, 400–412 (2013).
methyltransferase and is required for RNAi-mediated 59. Haag, J. R. et al. In vitro transcription activities of 79. Sienski, G., Donertas, D. & Brennecke, J.
heterochromatin formation. RNA Biol. 2, 106–111 Pol IV, Pol V, and RDR2 reveal coupling of Pol IV and Transcriptional silencing of transposons by PIWI and
(2005). RDR2 for dsRNA synthesis in plant RNA silencing. maelstrom and its impact on chromatin state and
35. Horn, P. J., Bastie, J. N. & Peterson, C. L. A Mol. Cell 48, 811–818 (2012). gene expression. Cell 151, 964–980 (2012).
Rik1‑associated, cullin-dependent E3 ubiquitin ligase 60. Ye, R. et al. Cytoplasmic assembly and selective This study was the first to show that, in
is essential for heterochromatin formation. Genes Dev. nuclear import of Arabidopsis Argonaute4/siRNA D. melanogaster, Piwi and piRNAs silence
19, 1705–1714 (2005). complexes. Mol. Cell 46, 859–870 (2012). transposons at the transcriptional level genome-wide.

82 | FEBRUARY 2015 | VOLUME 16 www.nature.com/reviews/genetics

© 2015 Macmillan Publishers Limited. All rights reserved


REVIEWS

80. Wang, S. H. & Elgin, S. C. Drosophila Piwi functions 103. Wedeles, C. J., Wu, M. Z. & Claycomb, J. M. 125. Lai, F. et al. Activating RNAs associate with Mediator
downstream of piRNA production mediating a Protection of germline gene expression by the to enhance chromatin architecture and transcription.
chromatin-based transposon silencing mechanism in C. elegans Argonaute CSR‑1. Dev. Cell 27, 664–671 Nature 494, 497–501 (2013).
female germ line. Proc. Natl Acad. Sci. USA 108, (2013). 126. Lam, M. T. et al. Rev–Erbs repress macrophage gene
21164–21169 (2011). References 102 and 103 demonstrate that the expression by inhibiting enhancer-directed
81. Brower-Toland, B. et al. Drosophila Piwi associates AGO protein CSR‑1 protects its targets from transcription. Nature 498, 511–515 (2013).
with chromatin and interacts directly with HP1a. repression by the PRG‑1–HRDE‑1 pathway. Thus, 127. Orom, U. A. & Shiekhattar, R. Long noncoding RNAs
Genes Dev. 21, 2300–2311 (2007). C. elegans uses small RNAs not only to mark usher in a new era in the biology of enhancers. Cell
82. Mohn, F., Sienski, G., Handler, D. & Brennecke, J. non-self RNAs but also to designate expressed 154, 1190–1193 (2013).
The Rhino–Deadlock–Cutoff complex licenses sequences as self. 128. Harigaya, Y. et al. Selective elimination of messenger
noncanonical transcription of dual-strand piRNA 104. Conine, C. C. et al. Argonautes promote male fertility RNA prevents an incidence of untimely meiosis.
clusters in Drosophila. Cell 157, 1364–1379 (2014). and provide a paternal memory of germline gene Nature 442, 45–50 (2006).
83. Le Thomas, A. et al. Transgenerationally inherited expression in C. elegans. Cell 155, 1532–1544 This study was the first to reveal that the mRNAs
piRNAs trigger piRNA biogenesis by changing the (2013). of meiosis-specific genes are targeted for
chromatin of piRNA clusters and inducing precursor 105. Yang, L., Froberg, J. E. & Lee, J. T. Long noncoding destruction during vegetative growth in S. pombe.
processing. Genes Dev. 28, 1667–1680 (2014). RNAs: fresh perspectives into the RNA world. The authors defined a cis-RNA motif (the DSR) and
84. Aravin, A. A., Hannon, G. J. & Brennecke, J. Trends Biochem. Sci. 39, 35–43 (2014). a trans-protein factor (Mmi1) that are required for
The Piwi–piRNA pathway provides an adaptive 106. Ferrari, F., Alekseyenko, A. A., Park, P. J. & this regulation.
defense in the transposon arms race. Science 318, Kuroda, M. I. Transcriptional control of a whole 129. Yamashita, A. et al. Hexanucleotide motifs mediate
761–764 (2007). chromosome: emerging models for dosage recruitment of the RNA elimination machinery
85. Aravin, A. A. et al. A piRNA pathway primed by compensation. Nature Struct. Mol. Biol. 21, 118–125 to silent meiotic genes. Open Biol. 2, 120014
individual transposons is linked to de novo DNA (2014). (2012).
methylation in mice. Mol. Cell 31, 785–799 (2008). 107. Kohlmaier, A. et al. A chromosomal memory triggered 130. Tashiro, S., Asano, T., Kanoh, J. & Ishikawa, F.
86. Carmell, M. A. et al. MIWI2 is essential for by Xist regulates histone methylation in X inactivation. Transcription-induced chromatin association of RNA
spermatogenesis and repression of transposons in the PLoS Biol. 2, E171 (2004). surveillance factors mediates facultative
mouse male germline. Dev. Cell 12, 503–514 (2007). 108. Plath, K. et al. Role of histone H3 lysine 27 heterochromatin formation in fission yeast. Genes
87. Kuramochi-Miyagawa, S. et al. DNA methylation of methylation in X inactivation. Science 300, 131–135 Cells 18, 327–339 (2013).
retrotransposon genes is regulated by PIWI family (2003). 131. Zofall, M. et al. RNA elimination machinery targeting
members MILI and MIWI2 in murine fetal testes. 109. Silva, J. et al. Establishment of histone H3 methylation meiotic mRNAs promotes facultative heterochromatin
Genes Dev. 22, 908–917 (2008). on the inactive X chromosome requires transient formation. Science 335, 96–100 (2012).
88. Watanabe, T. et al. Role for piRNAs and noncoding recruitment of Eed–Enx1 polycomb group complexes. This paper shows that the machinery responsible
RNA in de novo DNA methylation of the imprinted Dev. Cell 4, 481–495 (2003). for clearing meiotic mRNAs during vegetative
mouse Rasgrf1 locus. Science 332, 848–852 (2011). 110. Zhao, J., Sun, B. K., Erwin, J. A., Song, J. J. & Lee, J. T. growth also promotes the formation of
89. De Fazio, S. et al. The endonuclease activity of Mili Polycomb proteins targeted by a short repeat RNA to heterochromatin islands at the corresponding
fuels piRNA amplification that silences LINE1 the mouse X chromosome. Science 322, 750–756 meiotic gene loci.
elements. Nature 480, 259–263 (2011). (2008). 132. Sugiyama, T. & Sugioka-Sugiyama, R. Red1 promotes
90. de Vanssay, A. et al. Paramutation in Drosophila linked 111. Rinn, J. L. et al. Functional demarcation of active and the elimination of meiosis-specific mRNAs in
to emergence of a piRNA-producing locus. Nature silent chromatin domains in human HOX loci by vegetatively growing fission yeast. EMBO J. 30,
490, 112–115 (2012). noncoding RNAs. Cell 129, 1311–1323 (2007). 1027–1039 (2011).
This study reports the first example of paramutation 112. Pandey, R. R. et al. Kcnq1ot1 antisense noncoding 133. Lee, N. N. et al. Mtr4‑like protein coordinates
in animals and describes how a locus can become RNA mediates lineage-specific transcriptional silencing nuclear RNA processing for heterochromatin
included in the piRNA repertoire. Together with through chromatin-level regulation. Mol. Cell 32, assembly and for telomere maintenance. Cell 155,
reference 16, it illustrates how small RNAs can 232–246 (2008). 1061–1074 (2013).
serve as the carriers of epigenetic signals. 113. Maenner, S. et al. 2D structure of the A region of Xist 134. Shah, S., Wittmann, S., Kilchert, C. & Vasiljeva, L.
91. Chandler, V. L. Paramutation’s properties and puzzles. RNA and its implication for PRC2 association. PLoS LncRNA recruits RNAi and the exosome to
Science 330, 628–629 (2010). Biol. 8, e1000276 (2010). dynamically regulate pho1 expression in response
92. Yigit, E. et al. Analysis of the C. elegans Argonaute 114. da Rocha, S. T. et al. Jarid2 is implicated in the initial to phosphate levels in fission yeast. Genes Dev. 28,
family reveals that distinct Argonautes act sequentially Xist-induced targeting of PRC2 to the inactive X 231–244 (2014).
during RNAi. Cell 127, 747–757 (2006). chromosome. Mol. Cell 53, 301–316 (2014). 135. Egan, E. D., Braun, C. R., Gygi, S. P. & Moazed, D.
93. Guang, S. et al. An Argonaute transports siRNAs from 115. Brockdorff, N. Noncoding RNA and Polycomb Post-transcriptional regulation of meiotic genes by a
the cytoplasm to the nucleus. Science 321, 537–541 recruitment. RNA 19, 429–442 (2013). nuclear RNA silencing complex. RNA 20, 867–881
(2008). 116. Davidovich, C., Zheng, L., Goodrich, K. J. & Cech, T. R. (2014).
94. Burkhart, K. B. et al. A pre-mRNA-associating factor Promiscuous RNA binding by Polycomb repressive 136. Yamanaka, S., Yamashita, A., Harigaya, Y., Iwata, R. &
links endogenous siRNAs to chromatin regulation. complex 2. Nature Struct. Mol. Biol. 20, 1250–1257 Yamamoto, M. Importance of polyadenylation in the
PLoS Genet. 7, e1002249 (2011). (2013). selective elimination of meiotic mRNAs in growing
95. Burton, N. O., Burkhart, K. B. & Kennedy, S. 117. Kaneko, S., Son, J., Shen, S. S., Reinberg, D. & S. pombe cells. EMBO J. 29, 2173–2181 (2010).
Nuclear RNAi maintains heritable gene silencing in Bonasio, R. PRC2 binds active promoters and contacts 137. Duret, L., Chureau, C., Samain, S., Weissenbach, J. &
Caenorhabditis elegans. Proc. Natl Acad. Sci. USA nascent RNAs in embryonic stem cells. Nature Struct. Avner, P. The Xist RNA gene evolved in eutherians by
108, 19683–19688 (2011). Mol. Biol. 20, 1258–1264 (2013). pseudogenization of a protein-coding gene. Science
96. Ashe, A. et al. piRNAs can trigger a multigenerational 118. Kanhere, A. et al. Short RNAs are transcribed from 312, 1653–1655 (2006).
epigenetic memory in the germline of C. elegans. Cell repressed Polycomb target genes and interact 138. Gapp, K. et al. Implication of sperm RNAs in
150, 88–99 (2012). with Polycomb repressive complex‑2. Mol. Cell 38, transgenerational inheritance of the effects of
97. Buckley, B. A. et al. A nuclear Argonaute promotes 675–688 (2010). early trauma in mice. Nature Neurosci. 17,
multigenerational epigenetic inheritance and germline 119. Zhao, J. et al. Genome-wide identification of 667–669 (2014).
immortality. Nature 489, 447–451 (2012). Polycomb-associated RNAs by RIP-seq. Mol. Cell 40, 139. Rechavi, O. et al. Starvation-induced transgenerational
This study shows that HRDE‑1 is the AGO protein 939–953 (2010). inheritance of small RNAs in C. elegans. Cell 158,
responsible for transgenerational inheritance of 120. Kaneko, S. et al. Interactions between JARID2 and 277–287 (2014).
classical RNAi-mediated silencing in C. elegans and noncoding RNAs regulate PRC2 recruitment to 140. Castel, S. E. & Martienssen, R. A. RNA interference
that this involves H3K9 methylation of silenced chromatin. Mol. Cell 53, 290–300 (2014). in the nucleus: roles for small RNAs in transcription,
genes in the germ line. References 114 and 120 show that Jarid2 acts as epigenetics and beyond. Nature Rev. Genet. 14,
98. Bagijn, M. P. et al. Function, targets, and evolution of an essential intermediate in Xist-mediated 100–112 (2013).
Caenorhabditis elegans piRNAs. Science 337, recruitment of PRC2 to chromatin, and that Xist 141. Bernstein, E., Caudy, A. A., Hammond, S. M. &
574–578 (2012). repeat A is dispensable for PRC2 targeting. Hannon, G. J. Role for a bidentate ribonuclease in the
99. Lee, H. C. et al. C. elegans piRNAs mediate the 121. Kim, T. K. et al. Widespread transcription at neuronal initiation step of RNA interference. Nature 409,
genome-wide surveillance of germline transcripts. activity-regulated enhancers. Nature 465, 182–187 363–366 (2001).
Cell 150, 78–87 (2012). (2010). 142. Elbashir, S. M., Lendeckel, W. & Tuschl, T.
100. Shirayama, M. et al. piRNAs initiate an epigenetic 122. De Santa, F. et al. A large fraction of extragenic RNA RNA interference is mediated by 21- and
memory of nonself RNA in the C. elegans germline. Pol II transcription sites overlap enhancers. PLoS Biol. 22‑nucleotide RNAs. Genes Dev. 15, 188–200
Cell 150, 65–77 (2012). 8, e1000384 (2010). (2001).
Together with references 96, 98 and 99, this paper 123. Orom, U. A. et al. Long noncoding RNAs with 143. Zamore, P. D., Tuschl, T., Sharp, P. A. & Bartel, D. P.
defines the mechanism by which C. elegans piRNAs enhancer-like function in human cells. Cell 143, RNAi: double-stranded RNA directs the ATP-
recognize and target non-self RNAs for silencing, a 46–58 (2010). dependent cleavage of mRNA at 21 to 23 nucleotide
memory of which is then maintained by secondary This study reports the discovery of eRNAs, which intervals. Cell 101, 25–33 (2000).
siRNAs and the AGO protein HRDE‑1. are required to activate gene expression in cis. 144. Parrish, S., Fleenor, J., Xu, S., Mello, C. & Fire, A.
101. Claycomb, J. M. et al. The Argonaute CSR‑1 and its 124. Wang, K. C. et al. A long noncoding RNA maintains Functional anatomy of a dsRNA trigger: differential
22G‑RNA cofactors are required for holocentric active chromatin to coordinate homeotic gene requirement for the two trigger strands in RNA
chromosome segregation. Cell 139, 123–134 (2009). expression. Nature 472, 120–124 (2011). interference. Mol. Cell 6, 1077–1087 (2000).
102. Seth, M. et al. The C. elegans CSR‑1 argonaute This paper reports an eRNA that induces 145. Song, J. J. et al. The crystal structure of the
pathway counteracts epigenetic silencing to promote chromosomal looping and recruits H3K4 Argonaute2 PAZ domain reveals an RNA binding motif
germline gene expression. Dev. Cell 27, 656–663 methylation to activate transcription of several in RNAi effector complexes. Nature Struct. Biol. 10,
(2013). genes in the HOXA cluster. 1026–1032 (2003).

NATURE REVIEWS | GENETICS VOLUME 16 | FEBRUARY 2015 | 83

© 2015 Macmillan Publishers Limited. All rights reserved


REVIEWS

146. Wang, Y. et al. Nucleation, propagation and cleavage 155. Dalmay, T., Hamilton, A., Rudd, S., Angell, S. & 163. Halic, M. & Moazed, D. Dicer-independent primal
of target RNAs in Ago silencing complexes. Nature Baulcombe, D. C. An RNA-dependent RNA RNAs trigger RNAi and heterochromatin formation.
461, 754–761 (2009). polymerase gene in Arabidopsis is required Cell 140, 504–516 (2010).
147. Ma, J. B., Ye, K. & Patel, D. J. Structural basis for posttranscriptional gene silencing mediated 164. Yu, R., Jih, G., Iglesias, N. & Moazed, D.
for overhang-specific small interfering RNA by a transgene but not by a virus. Cell 101, Determinants of heterochromatic siRNA biogenesis
recognition by the PAZ domain. Nature 429, 543–553 (2000). and function. Mol. Cell 53, 262–276 (2014).
318–322 (2004). 156. Sijen, T. et al. On the role of RNA amplification 165. Kawakami, K., Hayashi, A., Nakayama, J. &
148. Ma, J. B. et al. Structural basis for 5ʹ‑end-specific in dsRNA-triggered gene silencing. Cell 107, Murakami, Y. A novel RNAi protein, Dsh1, assembles
recognition of guide RNA by the A. fulgidus Piwi 465–476 (2001). RNAi machinery on chromatin to amplify
protein. Nature 434, 666–670 (2005). 157. Stuwe, E., Toth, K. F. & Aravin, A. A. Small but heterochromatic siRNA. Genes Dev. 26, 1811–1824
149. Song, J. J., Smith, S. K., Hannon, G. J. & Joshua-Tor, L. sturdy: small RNAs in cellular memory and (2012).
Crystal structure of Argonaute and its implications epigenetics. Genes Dev. 28, 423–431 166. Marasovic, M., Zocco, M. & Halic, M. Argonaute and
for RISC slicer activity. Science 305, 1434–1437 (2014). Triman generate dicer-independent priRNAs and
(2004). 158. Khurana, J. S. & Theurkauf, W. piRNAs, transposon mature siRNAs to initiate heterochromatin formation.
150. Matranga, C., Tomari, Y., Shin, C., Bartel, D. P. & silencing, and Drosophila germline development. Mol. Cell 52, 173–183 (2013).
Zamore, P. D. Passenger-strand cleavage facilitates J. Cell Biol. 191, 905–913 (2010).
assembly of siRNA into Ago2‑containing RNAi enzyme 159. Senti, K. A. & Brennecke, J. The piRNA pathway: Acknowledgements
complexes. Cell 123, 607–620 (2005). a fly’s perspective on the guardian of the genome. D.H. was supported by the US National Science Foundation
151. Rand, T. A., Petersen, S., Du, F. & Wang, X. Trends Genet. 26, 499–509 (2010). Graduate Research Fellowship Program. Research in D.M.’s
Argonaute2 cleaves the anti-guide strand of siRNA 160. Ishizu, H., Nagao, A. & Siomi, H. Gatekeepers for laboratory is supported by grants from the US National
during RISC activation. Cell 123, 621–629 (2005). Piwi–piRNA complexes to enter the nucleus. Institutes of Health. D.M. is an Investigator of the Howard
152. Miyoshi, K., Tsukumo, H., Nagami, T., Siomi, H. & Curr. Opin. Genet. Dev. 21, 484–490 (2011). Hughes Medical Institute.
Siomi, M. C. Slicer function of Drosophila Argonautes 161. Brennecke, J. et al. Discrete small RNA-generating
and its involvement in RISC formation. Genes Dev. 19, loci as master regulators of transposon activity Competing interests statement
2837–2848 (2005). in Drosophila. Cell 128, 1089–1103 (2007). The authors declare no competing interests.
153. Buker, S. M. et al. Two different Argonaute complexes 162. Gunawardane, L. S. et al. A slicer-mediated
are required for siRNA generation and mechanism for repeat-associated siRNA 5ʹ end
heterochromatin assembly in fission yeast. Nature formation in Drosophila. Science 315, 1587–1590 SUPPLEMENTARY INFORMATION
Struct. Mol. Biol. 14, 200–207 (2007). (2007). See online article: S1 (figure)
154. Baulcombe, D. RNA silencing in plants. Nature 431, References 161 and 162 reported the discovery of ALL LINKS ARE ACTIVE IN THE ONLINE PDF
356–363 (2004). piRNA amplification by the ping-pong mechanism.

84 | FEBRUARY 2015 | VOLUME 16 www.nature.com/reviews/genetics

© 2015 Macmillan Publishers Limited. All rights reserved


Copyright of Nature Reviews Genetics is the property of Nature Publishing Group and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.